Noncarbapenems for the Treatment of Urinary Tract Infections Caused by Extended-Spectrum β-Lactamase-Producing Bacteria

C. Whitney White, PharmD, BCPS; Jeffrey A. Kyle, PharmD, BCPS; Crystal M. Deas, PharmD, BCPS; Jacob Campbell, PharmD


South Med J. 2019;112(8):438-443. 

In This Article


In this retrospective observational study from a community hospital, the use of noncarbapenems for the treatment of ESBL UTIs was not associated with increased LOS, higher rate of mortality, or longer duration of antibiotic therapy compared with the use of carbapenems. Noncarbapenem therapy was, however, associated with higher rates of microbiological failure, although differences were not considered statistically significant (P = 0.115).

Within the noncarbapenem cohort, higher rates of in vitro susceptibility did not necessarily correspond to better patient outcomes. Excluding carbapenems (100% susceptibility) and amikacin (100%, but only 8 isolates tested), E. coli isolates showed the highest in vitro susceptibility rates to nitrofurantoin (90%); however, nitrofurantoin was associated with the greatest number (2) and percentage (100%) of positive cultures repeated within 4 weeks of treatment (ie, microbiological failure).

Overall, our results are consistent with other studies that have examined the place of noncarbapenems in the treatment of ESBL UTIs. A 2013 retrospective cohort study of 42 episodes of ESBL UTIs in children in South Korea found no significant difference between outcomes in those treated with carbapenems versus those treated with noncarbapenems.[23] At least three retrospective studies have found aminoglycosides, particularly amikacin, to be viable alternatives to carbapenems in the treatment of ESBL UTIs.[24–26] In 2014, a 152-patient cohort study found noncarbapenems, which included aminoglycosides, β-lactam/β-lactamase inhibitors, fluoroquinolones, and trimethoprim/sulfamethoxazole, to be as effective as carbapenems in treating community-onset acute pyelonephritis caused by ESBL E. coli.[22] A 2008 case-control study with 122 patient cases found fosfomycin and amoxicillin/clavulanate to be clinically effective for ESBL cystitis.[4] It is important to note that the clinical success achieved with noncarbapenems in each of these studies was associated with agents that had proven in vitro activity.

Since the conclusion of this study, some additional studies have been published related to the treatment of ESBL infections. Results from trials with the newer antibiotics ceftazidime-avibactam and ceftolozane-tazobactam indicate that these agents may be effective alternatives in the treatment of ESBL UTIs.[27,28] Furthermore, a small retrospective study showed cefepime to be comparable to carbapenems for the treatment of ESBL UTIs.[29] In contrast, the MERINO study found that piperacillin-tazobactam was inferior to meropenem as a definitive therapy for ceftriaxone-nonsusceptible E. coli or K. pneumoniae bloodstream infections.[30]

The present study has several important limitations. First, the total cohort was relatively small (N = 50), and combining the results of agents in the noncarbapenem group limited the determination of specific efficacy compared with carbapenems. Second, UTIs were not differentiated or classified according to urinary tract location (eg, cystitis, pyelonephritis, prostatitis), and strict criteria were not used a priori to categorize the significance of the UTI. Third, although certain characteristics and measures, such as comorbidities and ICU admission, were useful in assessing baseline and hospitalization differences among patients, the severity of the UTI was not evaluated, nor was the overall severity of illness assessed by any standard measurement (eg, the Charlson Comorbidity Index). Because carbapenems are more likely to be initiated in patients who are more severely ill, this may serve as a confounding factor, which was not accounted for in our study. Fourth, clinical cure was not able to be assessed because of the lack of necessary data within the medical record and the patients who were lost to follow-up after discharge. LOS and mortality are more likely to be affected by confounding factors compared with the markers of clinical cure (eg, defervescence, symptom improvement). Although microbiological failure was recorded and analyzed when possible, repeated cultures were obtained for only 16% of carbapenem patients and 20% of noncarbapenem patients. Finally, the adverse effects of therapy were not assessed.

The strengths of the study include an even distribution of patients between cohorts, similar baseline characteristics between cohorts, and a sample size that is comparable to other related published studies.